7 Step AI Visibility Guide Behind 58k Mentions in 3 Months (4,231 citations Analyzed)

If you want AI to recommend you to buyers but

↳ You have no idea which AI to optimize for

↳ Everything you publish gets ignored

Then stop treating this as one job. It is five, and the sources barely overlap.

***Run this playbook manually or let this Claude SEO skill automate it 👈***

> I analyzed 100 buy-intent questions across five AI answer engines: ChatGPT, Gemini, Perplexity, Google AI Overview and Google AI Mode. The questions are 25 business software categories crossed with 4 question shapes, so no category or phrasing is over-weighted. That produced 500 answers carrying 4,231 citations to 1,318 unique sources, collected from the United States in English with no personalization. Sources were sorted into vendor, publisher, review directory, media, community and video. I hand-checked a stratified sample of 50 of the 1,318 and 47 matched, so read the vendor and publisher shares as accurate to a few points rather than to a decimal. This is one snapshot and it cannot show what caused any engine to pick any source.

TL;DR

↳ On 85 of 98 questions, all five surfaces cited zero sources in common

↳ ChatGPT and Gemini share nothing at all on a typical question. Their median overlap is 0.000

↳ Perplexity cites 17 sources per answer. ChatGPT and Gemini cite 4

↳ Reddit is the most cited source in the study, on all five surfaces, 219 answers

↳ ChatGPT cites Reddit in 0% of answers to "best X software" and 24.1% of answers to "recommendations from real users"

↳ YouTube is cited in 63 of 100 Google AI answers and in 0 of ChatGPT's and Perplexity's

↳ Vendor sites are the biggest single bucket everywhere, between 36.0% and 47.4%

↳ Step 1: pick the surface you are actually optimizing for

↳ Step 2: get listed and reviewed on G2 and Capterra

↳ Step 3: get named inside Reddit threads

↳ Step 4: put one demo video on YouTube if you care about Google

↳ Step 5: answer the question on your own site

↳ Step 6: get into the specific listicles each surface leans on

↳ Step 7: measure each surface separately, using the phrasings your buyers use

The five engines agree on almost nothing

I expected heavy overlap. Same web, same questions, same week.

On 85 of the 98 questions all five surfaces answered, there was not one source that all five cited. The median count of sources shared by all five is zero.

Pairwise it is just as thin. Median overlap between two surfaces on the same question, where 1.0 would mean identical sources:

ChatGPT and Gemini sit at 0.000. On a typical buying question they hand the user two entirely separate sets of sources. The closest pair in the whole study is Perplexity and Google AI Overview at 0.158, which still means roughly one source in six.

So "getting cited by AI" is not a thing you can do once.

They do not even cite the same amount

Perplexity cites 17 sources per answer and never dropped below 14. ChatGPT and Gemini both sit at 4. That alone changes your odds: on Perplexity there are four times as many slots to win.

Gemini is the tightest of all. Its widest answer in the whole study cited 6 sources, and one answer cited none.

The question shape changes who gets cited

This is the finding I did not predict.

Reddit is the single most cited source in the study, appearing in 219 answers

across all five surfaces. But the average hides what is actually happening.

ChatGPT cited Reddit in 0% of its citations for "best X software", 0% for "best X tool for a small business" and 0% for "cheapest X software that is actually good". Then for "X software recommendations from real users" it jumped to 24.1%. Gemini does the same thing, 0% to 19.5%.

Google's two surfaces do not behave this way. AI Overview cites Reddit at 6.5%, 4.6%, 6.4% and 8.4% across the four shapes, which is close to flat.

The reading: ChatGPT and Gemini decide what kind of source to fetch based on how you phrase the question, and Google's surfaces mostly do not. One caveat before you build a strategy on it. Four question shapes over 25 categories is a small grid, and the socialproof phrasing contains the words "real users", which may be doing the work rather than the buying intent behind it.

YouTube is a Google thing and nothing else

YouTube appears in 63 of 100 AI Overview answers and 63 of 100 AI Mode answers. ChatGPT cited it zero times in 100 answers. Perplexity cited it zero times. Gemini managed 3.

Google owns YouTube, so the direction is no surprise, though the size of it is. A demo video does nothing at all for your ChatGPT citations, while on Google's two surfaces it is one of the biggest levers in the whole study.

What each surface actually leans on

Vendor sites are the largest bucket on every surface, from 36.0% on AI Overview to 47.4% on AI Mode. Your own pages are still the thing most likely to be quoted back at a buyer.

After that the surfaces split. ChatGPT leans on review directories at 13.5%, roughly double everyone except Perplexity. Perplexity leans on media at 15.2%. Google AI Mode leans on independent publishers at 23.6% and sends 45.4% of its citations to sources no other surface cited once.

Two mechanics worth knowing before you try to measure any of this. Google returns every citation as a google.com/goto redirect with an opaque payload, so the real publisher only survives in a separate field. ChatGPT appends utm_source=chatgpt.com to the links it hands out, which means you can already count ChatGPT referrals in your own analytics today.

The 7-step playbook

Step 1: Pick the AI you are optimizing for

You cannot work all five at once and the overlap data says you would not benefit from trying.

I mean, you can, just do it one step at a time.

↳ Check your analytics for utm_source=chatgpt. com and for perplexity. ai referrals. Those are countable today

↳ Ask your last 20 customers which assistant they used before they found you

↳ Pick one surface and one backup, then run steps 2 to 6 against those

Time: 30 minutes.

Step 2: Get listed and reviewed on G2 and Capterra

Review directories are the only third-party category cited on all five surfaces. G2 is the second most cited source in the whole study at 161 answers, behind only Reddit.

↳ G2 is cited in 70 of Perplexity's 100 answers, 32 of AI Mode's, 26 of ChatGPT's, 23 of AI Overview's and 10 of Gemini's

↳ Capterra is cited in 34 of Perplexity's and 22 of ChatGPT's

↳ Gartner, Software Advice and TechnologyAdvice all appear on 4 or 5 surfaces

Claim the profile, then get real reviews on it. This is the highest-coverage

single action in the study.

Step 3: Get named inside Reddit threads

— "Reddit Radar scanning for threads on Distribb.io (beta)"

Reddit is cited more than any other source (on some AI's, consider ChatGPT dropped it last week).

Do not spam it. They're too good at spotting spam.

↳ Find threads asking for recommendations in your category

↳ Answer as yourself, disclose that you built the thing, and name competitors honestly

↳ Expect this to pay off mostly on Perplexity and Google, which cite Reddit whatever the phrasing

On Gemini your Reddit mentions only surface when the buyer asks in the "real users" shape, so treat it as coverage for one question type rather than a general win.

Step 4: Put one demo video on YouTube

— "Done for you video distribution in Distribb.io targeting buying queries"

63 of 100 Google AI answers cited YouTube. Skip this entirely if step 1 pointed you at ChatGPT or Perplexity, where it scored 0 of 100.

↳ One video per category term you want to be named in

↳ Show the product finishing a real job, since that is what the cited videos do

↳ Title it as the question, not as your product name

Step 5: Answer the buying question on your own site

— "FAQ inclusion in Distribb.io articles targeting buying questions"

Vendor pages are the biggest bucket on all five surfaces, so this is the one step that pays on every one of them.

↳ One page per buying question, in the words a buyer would type

↳ State pricing on the page. "Cheapest X that is actually good" was one of the four question shapes and every surface answered it

↳ Keep a comparison page naming your real competitors

Step 6: Get into the listicles your chosen AI cites

— "Link outreach feature on Distribb.io to automate listicle visibility"

Independent publishers are between 17.0% and 30.8% of citations. The specific sites differ by surface, so pull them from your own scan rather than copying mine.

↳ From this study: PCMag, TechRadar, Forbes, Fit Small Business, Venture Harbour and Zapier's blog all appear on 3 or more surfaces

↳ Zapier is worth singling out. It is a software vendor whose listicle blog was cited 128 times across all five surfaces, which is what a vendor-run publication can be worth

↳ Pitch the writer with a product account and a real differentiator, not a press release

Step 7: Measure each AI separately

— "AI visibility tracking in Distribb.io"

An aggregate "AI visibility" number will average away everything in this study.

↳ Track citations per surface, never pooled

↳ Use several question shapes per category, because step 3 showed phrasing moves the source mix by 24 points

↳ Re-scan monthly. This is one snapshot and these surfaces change their retrieval without telling anyone

Recap: run the seven steps in order

↳ Step 1: Pick the surface you are optimizing for, using chatgpt.com referrals and what customers tell you

↳ Step 2: Claim and build reviews on G2 and Capterra, the only third party cited on all five

↳ Step 3: Get named inside Reddit threads, the most cited source in the study

↳ Step 4: Publish one YouTube demo per category term, but only if your buyers use Google

↳ Step 5: Answer each buying question on your own site, since vendor pages lead on every surface

↳ Step 6: Get into the listicles your chosen surface actually leans on

↳ Step 7: Measure per surface and per question shape, monthly

Five engines, 100 identical questions, and on 85 of them not a single source in common.

**Try Distribb free and track which AI surfaces name you 👈**

Thanks for reading.

Borja